R Notebook

sat <- read.csv("train.csv")
head(sat)

serie <- sat[sat$sat_id==1,]
head(serie)

serie$epoch <- as.numeric(as.POSIXct(serie$epoch))
head(serie$epoch)

[1] 1388530800 1388530800 1388530800 1388530800 1388530800
[6] 1388530800

rm.list <- c()
for (i in 2:length(serie$epoch)){
  if (serie$epoch[i-1] == serie$epoch[i]){
    rm.list <- c(rm.list, i)
  }
}

seriex <- ts(serie$x, freq=4)
plot(seriex)

seriex.d24 <- diff(seriex, 4)
plot(seriex.d24)
abline(h=0, lty=2)

seriex.d24.d24 <- diff(seriex.d24, 4)
plot(seriex.d24.d24)
abline(h=0, lty=2)

seriex.d24.d24.d24 <- diff(seriex.d24.d24,4)
plot(seriex.d24.d24.d24)
abline(h=0, lty=2)

var(seriex)

[1] 198348124

var(seriex.d24)

[1] 585124254

var(seriex.d24.d24)

[1] 1680495618

var(seriex.d24.d24.d24)

[1] 4760606473

seriex.d24.d24.d1 <- diff(seriex.d24.d24)
plot(seriex.d24.d24.d1)
abline(h=0, lty=2)

var(seriex.d24.d24)

[1] 1680495618

var(seriex.d24.d24.d1)

[1] 2061644557

var(diff(seriex.d24.d24.d1))

[1] 2765817130

par(mfrow=c(1,2))
acf(seriex.d24.d24, col=c(2,rep(1,3)), lag=100)
pacf(seriex.d24.d24, col=c(rep(1,3),2), lag=100)

par(mfrow=c(1,2))
acf(seriex.d24.d24, col=c(2,rep(1,23)), lag=100)
pacf(seriex.d24.d24, col=c(rep(1,23),2), lag=100)

mod <- arima(seriex, order=c(2,0,0), seasonal=list(order=c(3,2,0), frequency=4))

res <- resid(mod)

plot(res)

qqnorm(res)
qqline(res)

hist(res, breaks=40, freq=F)

tsdiag(mod,40)

pred <- predict(mod, n.ahead=12)

pre <- pred$pred

lo <- pre - 1.96 * pred$se
up <- pre + 1.96 * pred$se

ts.plot(seriex,lo,up,pre,lty=c(1,2,2,1),col=c(1,4,4,2),xlim=c(0,12), ylim=c(min(seriex)-1000, max(seriex)+100000), type="o") 
abline(v=0:12,lty=3,col=4)

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